Use of repurposed and adjuvant drugs in hospital patients with covid-19: multinational network cohort study
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
RESEARCH
Use of repurposed and adjuvant drugs in hospital patients with
covid-19: multinational network cohort study
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
Albert Prats-Uribe,1 Anthony G Sena,2,3 Lana Yin Hui Lai,4 Waheed-Ul-Rahman Ahmed,5,6
Heba Alghoul,7 Osaid Alser,8 Thamir M Alshammari,9 Carlos Areia,10 William Carter,11
Paula Casajust,12 Dalia Dawoud,13,14 Asieh Golozar,15,16 Jitendra Jonnagaddala,17
Paras P Mehta,18 Mengchun Gong,19 Daniel R Morales,20,21 Fredrik Nyberg,22 Jose D Posada,23
Martina Recalde,24,25 Elena Roel,24,25 Karishma Shah,5 Nigam H Shah,23 Lisa M Schilling,11
Vignesh Subbian,26 David Vizcaya,27 Lin Zhang,28,29 Ying Zhang,19 Hong Zhu,30 Li Liu,30
Jaehyeong Cho,31 Kristine E Lynch,32 Michael E Matheny,33,34 Seng Chan You,35
Peter R Rijnbeek,3 George Hripcsak,36 Jennifer CE Lane,5 Edward Burn,1,24 Christian Reich,37
Marc A Suchard,38 Talita Duarte-Salles,24 Kristin Kostka,37,39 Patrick B Ryan,2,40
Daniel Prieto-Alhambra1
For numbered affiliations see Abstract in Spain), azithromycin (from 15 (4.9%) in China to
end of the article. Objective 1473 (57.9%) in Spain), combined lopinavir and
Correspondence to: P B Ryan To investigate the use of repurposed and adjuvant ritonavir (from 156 (RESEARCH
ventilation, and duration of hospital stay.5 Other database from February to August 2020), Optum (Eden
drugs, such as interferon and lopinavir combined with Prairie, MN) deidentified electronic health record
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
ritonavir have also been shown to be ineffective.5 7 dataset (Optum-EHR, February to October 2020),
In the absence of approved antivirals for the Tufts Medical Center Clinical Academic Research
treatment of covid-19, the cornerstone of management Enterprise Trust (Tufts Research Data Warehouse
has been supportive care, with adjunctive treatments (TRDW), February to May 2020), and the Department
playing a major role. The two recognised drug classes of Veterans Affairs (VA-OMOP, February to June
used for adjunctive treatment are corticosteroids and 2020). Data for South Korea came from nationwide
anticytokines (eg, tocilizumab). A large randomised claims recorded in the Health Insurance Review and
controlled trial and meta-analysis showed that the Assessment (HIRA, February to April 2020).21 Inpatient
glucocorticosteroid dexamethasone and corticoste electronic health record data from Spain was obtained
roids reduced mortality among patients receiving from HM Hospitales (March to April 2020) and
mechanical ventilation or oxygen.8 9 Tocilizumab was Hospital del Mar (February to August 2020). Data from
found to significantly reduce mortality in patients China was extracted from nine hospitals in Honghu,
admitted to hospital with covid-19.10 Although supported by Nanfang Hospital and Southern Medical
additional adjunctive treatments are recognised in University, and contained full electronic health record
2020 guidelines, including antithrombotics, statins, data (NFHCRD database, January to April 2020). Data
and antihypertensives,11-15 recommendations for on drug use in patients receiving intensive care were
covid-19 treatment in clinical guidelines have varied available from IQVIA Hospital CDM, Premier, Optum-
both geographically and temporally.16 EHR, VA-OMOP, HM Hospitales, and Hospital del Mar.
Regulators and public health agencies need to keep Supplementary table 1 provides a detailed description
up with trends in covid-19 clinical practice. Tweets of the databases.
and press conferences have been shown to influence
entire practice patterns but based on little evidence Study participants
for the utility of treatments. Although attention has Patients admitted to hospital with a recorded diagnosis
shifted to vaccine surveillance since December 2020, of covid-19 or a positive polymerase chain reaction test
there is still a need to understand what treatments are result for SARS-CoV-2 between January and December
effective for individual patients and at what harm This 2020 were included. A second cohort of patients who
body of evidence is critical for comparative purposes received intensive care was identified as a subset of
as more data become available during the pandemic. the former, defined by the initiation of mechanical
With known problems in the supply chain for certain ventilation, extracorporeal membrane oxygenation,
drugs, an understanding of what drugs are being used or tracheostomy. Index dates for the two cohorts were
to treat covid-19 at different stages of the disease could the date of admission to hospital and the date intensive
help resource constrained environments. care started, respectively.
We investigated the use of repurposed and adjunctive
drugs among patients admitted to hospital with Drugs of interest
covid-19 and among patients receiving intensive care We obtained information on all drugs prescribed or
in the United States, South Korea, Spain, and China. dispensed during hospital admission. For the study
of treatments used for covid-19, we assessed all drugs
Methods included in at least two randomised controlled trials
This multinational network cohort study was based on according to the covid-19 clinical trial tracker.22 The
hospital electronic health records and claims data. We resulting list was circulated to stakeholders with a role
mapped data from different sites to the Observational in drug development and research (eg, key opinion
Medical Outcomes Partnership (OMOP) Common Data leaders, pharmaceutical industry) and drug regulatory
Model (CDM).17 This approach allowed contributing agencies. All their suggestions were added to the final list
centres to execute analytical code in a distributed or of medicines under study. We classified the drugs into
federated fashion, where each site runs the analyses two groups: repurposed drugs—those with alternative
separately in-house and returns a results dataset indications but thought to be efficacious as antivirals;
without sharing patient level data. The study protocol and adjuvant drugs—those used to treat pneumonia
and analytical package were released on 11 June 2020, or prevent or treat complications from covid-19.23
and iterative updates are continually released through Supplementary table 3 lists the drugs considered.
GitHub.18 Our study was also published as a preprint.19 For the main results, we focused on drugs covered in
the living World Health Organization guideline for
Data sources drugs—hydroxychloroquine, lopinavir combined with
Data were obtained from the US, South Korea, Spain, ritonavir, remdesivir, and dexamethasone.7
and China. Electronic health record data from the US
were obtained from Columbia University Irving Medical Statistical analysis
Center (CUIMC, February to December 2020), IQVIA We summarise age, sex, and history of medical
Hospital CDM (February to October 2020), STAnford conditions as proportions (the number of participants
medicine Research data Repository (STARR-OMOP within a category, divided by the total number of
database20 from February to May 2020, and Premier participants). Supplementary table 2 shows the
2 doi: 10.1136/bmj.n1038 | BMJ 2021;373:n1038 | the bmjRESEARCH
clinical codes and time windows used to identify supplementary appendix provides time series graphs
medical conditions. for all drugs and groupings.
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
Drug use was calculated from the index date
(admission date or initiation of intensive care) to 30 Patient and public involvement
days after, or discharge, or death, when these dates No funding was available for patient or public
were available in the database. We calculated use for involvement in this project. Urgency because of the
each drug and major drug class. Prevalence of drug use covid-19 pandemic and restrictions also prevented
was determined by the proportion of participants with us from actively involving patients, although the
any active prescription or dispensation of a certain Observational Health Data Sciences and Informatics
drug or drug during hospital admission or the period community welcomes members of the public to engage
of intensive care. Figure 1 provides a timeline of the with its work. No patients were involved in setting the
study. Supplementary figure 1 shows which drugs research question or the outcome measures. Patients
could potentially have been prescribed in the month were not invited to comment on the study design, not
before hospital admission. consulted to develop patient relevant outcomes or
All drugs and additional time windows (previous interpret the results, and not invited to contribute to
year, previous month, and on index date) are the writing or editing of this document for readability
reported in full and will be updated in a dedica or accuracy.
ted interactive website (https://data.ohdsi.org/
Covid19CharacterizationCharybdis/) as more data Results
become available. All (aggregated) data can be A total of 303 264 patients identified from 11 databases
downloaded from this website. were included: 290 131 participants were from the US
To better visualise drug use, we generated rainbow (744 from California, 326 from Massachusetts, 7353
plots for each database. These plots display the from New York, 10 951 from US-wide Veterans Affairs,
proportion of users of each drug using Anatomical and 270 757 from US-wide databases: Premier, IQVIA
Therapeutic Chemical groupings. We also created Hospital CDM, and Optum-EHR), 7599 from South
lollipop plots of drug use to show the heterogeneity Korea, 5230 from Spain, and 304 from China, Of these
for all selected repurposed and adjuvant drugs 303 264 participants, 62 963 (from VA-OMOP, Premier,
(see supplementary file). On the basis of drug use Optum-EHR, IQVIA Hospital CDM, Hospital del Mar,
proportions, we determined the top five most used and HM Hospitales) received intensive care.
repurposed drugs and top 10 most used adjuvant The results of this study are available in an
drugs for each database and setting; use of the focused interactive website (https://data.ohdsi.org/
medicines are depicted by gauge plots. Covid19CharacterizationCharybdis/). This website
We calculated use of the selected drugs by month of contains both the summary results presented here and
index date (admission to hospital or start of intensive further details, including all drugs and comorbidities
care). To ensure enough time points, we selected recorded for the two cohorts.
databases with two or more months of data available Table 1 presents the baseline characteristics of
for each drug. Drug use was plotted for each calendar the patients admitted to hospital with covid-19.
month in the study period. A timeline of selected Supplementary table 4 shows the results for patients
relevant events, such as regulatory decisions or trial who received intensive care. Age varied slightly across
results for the selected medicines, was added. The data sources, but most patients were within the age
End of follow-up,
Admission 30 days, or death,
No admission to to hospital or discharge
hospital for covid-19
in previous six months Positive test result for, or diagnosis of, covid-19
Active drug use on patients admitted to hospital
Baseline covariates Mechanical ventilation, End of follow-up,
extracorporeal membrane 30 days, or death,
Baseline drug use oxygenation, or tracheostomy or discharge
Active drug use on those requiring intensive care
Days from D-30 D-21 D-1 D0 D30
index date
Days from Intensive Intensive
admission to care0 care30
intensive care
Fig 1 | Timeline of study
the bmj | BMJ 2021;373:n1038 | doi: 10.1136/bmj.n1038 3RESEARCH
Table 1 | Baseline characteristics of patients admitted to hospital with covid-19, stratified by data source. Data are numbers (percentages)
South Spain: HM Spain: USA: IQVIA USA:
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
USA: Korea: Hospitales Hospital Hospital China: USA: USA: STARR- USA: USA:
CUIMC HIRA (Spain; del Mar CDM NFHCRD Optum-EHR Premier OMOP TRDW VA-OMOP
Characteristics (n=7353) (n=7599) n=2544) (n=2686) (n=77 853) (n=304) (n=36 717) (n=156 187) (n=744) (n=326) (n=10 951)
Sex:
Female 3897 (53) 4483 (59) 1043 (41) 1424 (53) 38 139 (49) 149 (49) 18 359 (50) 74 970 (48) 372 (50) 140 (43) 767 (7)
Age (years):
0-4 221 (3) 76 (1) 25 (1) 13 (0.5) 2335 (3)RESEARCH
CUIMC (n=7353)
100
Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%) Prevalence (%)
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
50
0
HIRA (n=7599)
100
50
0
HM Hospitales (n=2544)
100
50
0
Hospital del Mar (n=2686)
100
50
0
IQVIA hospital CDM (n=77 853)
100
50
0
NFHCRD (n=304)
100
50
0
Optum-EHR (n=36 717)
100
50
0
Premier (n=156 187)
100
50
0
STARR-OMOP (n=744)
100
50
0
TRDW (n=326)
100
50
0
VA-OMOP (n=10 951)
100
50
0
Alimentary tract and metabolism Musculoskeletal system Drug use
Anti-infectives for systemic use Nervous system
Antiparasitic products, insecticides, and repellents Respiratory system
Blood and blood forming organs Sensory organs
Cardiovascular system Systemic hormonal preparations, excluding sex hormones and insulins
Dermatologicals Uncategorised
Genitourinary system and sex hormones Various
Fig 2 | Percentage of any use (one day or more) of all drugs (rainbow plot) on days 0 to 30 after hospital admission in patients with a positive
test result for or diagnosis of covid-19. CUIMC=Columbia University Irving Medical Center; HIRA=Health Insurance Review and Assessment;
OMOP=Observational Medical Outcomes Partnership; Optum-EHR=Optum deidentified electronic health record dataset; STARR=STAnford medicine
Research data Repository; TRDW=Tufts Research Data Warehouse; VA=Veterans Affairs
the bmj | BMJ 2021;373:n1038 | doi: 10.1136/bmj.n1038 5RESEARCH
Table 2 | Top five most used repurposed drugs in each data source in patients admitted to hospital with covid-19 on days 0 to 30 after hospital
admission. Data are treatment (percentage of patients admitted to hospital who received the medicine)
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
Spain: HM Spain: USA: IQVIA
USA: South Korea: Hospitales Hospital Hospital China: USA: USA: USA:
CUIMC HIRA (Spain; del Mar CDM NFHCRD Optum-EHR USA: Premier STARR-OMOP USA: TRDW VA-OMOP
Ranking (n=7353) (n=7599) n=2544) (n=2686) (n=77 853) (n=304) (n=36 717) (n=156 187) (n=744) (n=326) (n=10 951)
1 HCQ (22.3) L-R (34.9) HCQ (85.1) HCQ (40.2) AZM (47.4) UMF (78.3) AZM (37.0) AZM (46.6) AZM (8.7) AZM (28.5) AZM (33.0)
2 AZM (21.4) HCQ (27.4) AZM (57.9) AZM (7.9) HCQ (9.3) RBV (21.1) HCQ (20.5) HCQ (22.8) HCQ (3.2) HCQ (19.9) HCQ (13.5)
3 RMD (7.7) AZM (13.7) L-R (50.5) L-R (4.4) RMD (7.3) OST (13.2) L-R (1.2) OST (0.5) OST (1.1) OST (RESEARCH
China South Spain USA
Korea
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
NFHCRD HIRA HM Hospital CUIMC IQVIA Optum-EHR Premier STARR-OMOP TRDW VA-OMOP
Jan-Apr Feb-Apr Hospitales del Mar Feb-Dec hospital CDM Feb-Oct Feb-Aug Feb-May Feb-May Feb-Jun
(n=304) (n=7599) Mar-Apr Feb-May (n=7353) Feb-Oct (n=36 717) (n=156 187) (n=744) (n=326) (n=10 951)
(n=2544) (n=2686) (n=77 853)
Hydroxychloroquine
Dexamethasone
Azithromycin
Tocilizumab
Ritonavir
Remdesivir
Fig 3 | Percentage of any use (one day or more) of selected drugs on days 0 to 30 after hospital admission in patients with a positive test result for
or diagnosis of covid-19. CUIMC=Columbia University Irving Medical Center; HIRA=Health Insurance Review and Assessment; OMOP=Observational
Medical Outcomes Partnership; Optum-EHR=Optum deidentified electronic health record dataset; STARR=STAnford medicine Research data
Repository; TRDW=Tufts Research Data Warehouse; VA=Veterans Affairs
at the start of the pandemic in South Korea and Spain, Discussion
with a downward trend over time. Remdesivir was only This study reports on the use of repurposed and
recorded in CUIMC and IQVIA Hospital CDM, and it adjunctive drugs for the treatment of patients admitted
showed a slight upward trend from June onwards. to hospital with covid-19, including those who received
February 2020
4 February Wang et al - Remdesivir and chloroquine effectively inhibit covid-19 in vitro
March 2020
9 March Yao et al - Hydroxychloroquine shows superior in vitro activity to chloroquine
19 March President Trump promotes hydroxychloroquine in press conference
20 March Gautret et al - Open-label non-randomized clinical trial shows effectiveness
28 March US Food and Drug Administration issues an emergency use authorisation
31 March Chen et al - Preprint of a randomised controlled trial suggests that hydroxychloroquine reduces time to clinical recovery
April 2020
10 April Lane et al - Observational data show that azithromycin combined with hydroxychloroquine may increase
cardiovascular mortality
24 April FDA and European Medicines Agency caution against the use of hydroxychloroquine owing to potential
heart rhythm problems
May 2020
7 May Geleris et al - Lack of effectiveness of hydroxychloroquine on observational data
28 May WHO halts hydroxychloroquine arm of Solidarity trial
June 2020
8 June Recovery trial press note shows that hydroxychloroquine has no effect on
covid-19
15 June FDA revokes emergency use ruling for hydroxychloroquine
100
Percentage
South Korea USA
80 HIRA (n=7599) IQVIA Hospital CDM (n=77 853)
Spain CUIMC (n=7353)
60 HM Hospitales (n=2544) Optum-EHR (n=36 717)
Hospital del Mar (n=2686) Premier (n=156 187)
40 STARR-OMOP (n=744)
VA-OMOP (n=10 951)
20
0
February March April May June July August September October November December
Fig 4 | Time trends in hydroxychloroquine use on days 0 to 30 after hospital admission in patients with a positive test result for or diagnosis of
covid-19 by month. CUIMC=Columbia University Irving Medical Center; HIRA=Health Insurance Review and Assessment; OMOP=Observational
Medical Outcomes Partnership; Optum-EHR=Optum deidentified electronic health record dataset; STARR=STAnford medicine Research data
Repository; VA=Veterans Affairs
the bmj | BMJ 2021;373:n1038 | doi: 10.1136/bmj.n1038 7RESEARCH
February 2020
4 February Wang et al - Remdesivir and chloroquine effectively inhibit covid-19 in vitro
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
May 2020
22 May Beigel et al - ACTT-1 trial shows remdesivir shortens time to recovery in patients admitted to hospital
October 2020
15 Oct Solidarity trial preprint shows no effect of
remdesivir on mortality, ventilation, or
hospital stay in patients admitted to hospital
25
Percentage
USA
20 IQVIA Hospital CDM (n=77 853)
CUIMC (n=7353)
15
10
5
0
February March April May June July August September October November December
Fig 5 | Time trends in remdesivir use on days 0 to 30 after hospital admission in patients with a positive test result for or diagnosis of covid-19 by
month. ACTT-1=Adaptive COVID-19 Treatment Trial 1; CUIMC=Columbia University Irving Medical Center
intensive care, as recorded in electronic medical from covid-19, including antibiotics, anticoagulants,
records and claims data across three continents. A total corticosteroids, vitamin D supplements, and, to a
of 303 264 patients were admitted to hospital of whom lesser degree, antihypertensives, antacids, statins,
62 963 received intensive care for covid-19 in the US, and metformin. The use of adjunctive drugs increased
South Korea, Spain, and China. among patients who required intensive care.
We observed high heterogeneity in the use of Hydroxychloroquine has been given much publicity
repurposed drugs, with great variability in the use since the start of the pandemic. Its use has been
of hydroxychloroquine both geographically and supported or endorsed on the basis of misleading
temporally. Similar trends were observed for azithro evidence from flawed but heavily publicised
mycin. Use of antiretrovirals also varied greatly, with studies.24-26 Numerous randomised controlled trials
lopinavir-ritonavir use ranging from 0% in the US (VA- have, however, shown no benefit. The Recovery
OMOP) to 35% in South Korea (HIRA), and highest at trial of 1542 hospital patients with covid-19 treated
50% in Spain (HM Hospitales). with hydroxychloroquine showed no effects on 28
Adjunctive treatments have been extensively used day mortality compared with usual care.27 Another
for the prevention of or treatment for complications randomised controlled trial studied the efficacy of
June 2020
29 June Recovery trial press note shows no clinical benefit from the
use of lopinavir-ritonavir in patients admitted to hospital
October 2020
15 Oct Solidarity trial preprint shows no effect of
lopinavir-ritonavir on mortality, ventilation,
or hospital stay in patients admitted to hospital
75
Percentage
South Korea USA
60 HIRA (n=7599) IQVIA Hospital CDM (n=77 853)
Spain CUIMC (n=7353)
45 HM Hospitales (n=2544) Optum-EHR (n=36 717)
Hospital del Mar (n=2686) Premier (n=156 187)
30 VA-OMOP (n=10 951)
15
0
February March April May June July August September October November December
Fig 6 | Time trends combined lopinavir and ritonavir use on days 0 to 30 after hospital admission in patients with a positive test result for
or diagnosis of covid-19 by month. CUIMC=Columbia University Irving Medical Center; HIRA=Health Insurance Review and Assessment;
OMOP=Observational Medical Outcomes Partnership; Optum-EHR=Optum deidentified electronic health record dataset; VA=Veterans Affairs
8 doi: 10.1136/bmj.n1038 | BMJ 2021;373:n1038 | the bmjRESEARCH
South Korea USA
HIRA (n=7599) IQVIA Hospital CDM (n=77 853)
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
Spain CUIMC (n=7353)
HM Hospitales (n=2544) Optum-EHR (n=36 717)
Hospital del Mar (n=2686) Premier (n=156 187) June 2020
STARR-OMOP (n=744) 16 June Recovery trial press note shows that dexamethasome reduces death by
up to one third in patients admitted to hospital with severe respiratory
VA-OMOP (n=10 951) disease
100
Percentage
80
60
40
20
0
February March April May June July August September October November December
Fig 7 | Time trends in dexamethasone use on days 0 to 30 after hospital admission in patients with a positive test result for or diagnosis of covid-19
by month. CUIMC=Columbia University Irving Medical Center; HIRA=Health Insurance Review and Assessment; OMOP=Observational Medical
Outcomes Partnership; Optum-EHR=Optum deidentified electronic health record dataset; STARR=STAnford medicine Research data Repository;
VA=Veterans Affairs
hydroxychloroquine as post-exposure prophylaxis patients. Umifenovir in China was the most prescribed
in 821 asymptomatic participants but was found not repurposed drug, consistent with Chinese guidelines
to prevent covid-19 illness after high or moderate and research.34 35
exposure to covid-19.28 Hydroxychloroquine use Adjunctive drugs used to prevent covid-19 or treat
increased rapidly when these studies appeared and complications differed noticeably worldwide. Use of
were heavily publicised and politically endorsed. corticosteroids ranged from about 10% of admitted
During March and April 2020, more than 50% of patients in South Korea (HIRA) to 67% of patients in
patients admitted to hospital with covid-19 were Stanford (California, US).
prescribed hydroxychloroquine. After several papers Before results were available from the Recovery trial,
and regulatory agencies warned about potential side there was a wide debate on whether corticosteroids
effects, especially when hydroxychloroquine was have a role in mitigating inflammatory organ
combined with azithromycin, the use of hydroxy injury.36 37 Most clinical guidelines did not recommend
chloroquine began to decline. Finally, after the the use of corticosteroids to treat covid-19,31 with
Solidarity trial halted its hydroxychloroquine arm and notable exceptions.24 38 The use of dexamethasone
the Recovery trial presented definitive evidence against was low in almost all settings in our study until June
the use of hydroxychloroquine, the FDA revoked its 2020, when the Recovery trial showed its efficacy in
approval for emergency use and prescribing decreased reducing death in patients admitted to hospital with
to almost 0% in all settings.29 30 severe covid-19 related disease receiving respiratory
We found that azithromycin, a macrolide antibiotic support.39 Corticosteroid use in general appeared to
with alleged antiviral efficacy against covid-19, was increase slowly during the study period.
also widely prescribed. Although several guidelines in The use of anticoagulants in our study was higher
2020 recommended the use of empirical antimicrobial than expected. Heparin use was widely prescribed
treatment, not all advocated its use.31 In mid-December in the US and Spain, but not in China or South
the Recovery trial showed no benefit from azithromycin Korea. Severe covid-19 has been associated with a
in patients admitted to hospital with covid-19.32 We coagulopathy, which when untreated leads to poor
were not able to see the impact in trends as we only clinical outcomes.40 Although several randomised
had data until December 2020. controlled trials are ongoing to evaluate the value of
Combined use of the protease inhibitors lopinavir anticoagulation in patients with covid-19, interim
and ritonavir was high in South Korea and Spain, with guidelines recommend the use of anticoagulants
the other databases showing a much lower use. This for prophylaxis against thromboembolism.12 41 The
was consistent with Korean and Spanish guidelines at use of antibiotics also varied widely, as did the use
the time of our study, which recommended protease of statins. Traditional Chinese medicines were not
inhibitors as antiviral treatments,11 12 probably based widely prescribed (RESEARCH
endorsements are aligned with changes in practice settings during 2020. Our study cohorts included both
patterns and potentially influence the decisions of academic hospitals (eg, at Columbia University and
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
regulators.42 Retrospective evaluation of management Stanford University) as well as nationwide data sources
and treatment practices during the pandemic are and including other less specialised treatment centres
necessary43 to safeguard against the increase in use of (eg, HIRA, IQVIA Hospital CDM). Owing to the varied
unproven and potentially harmful treatments, during settings we decided not to provide drug use by country
future waves of the pandemic and public health crises. or overall because it would not be representative of the
underlying populations.
Limitations of this study
Our study was based on routinely collected real world Conclusions
data (electronic health records and claims data), with Great interest has been shown in the safety and efficacy
the potential for misclassification of disease and of drugs used to treat covid-19, but little evidence
treatments. We only included patients with a clinical exists on the prescribing patterns for repurposed and
diagnosis of covid-19 or a positive polymerase chain adjuvant drugs in routine clinical practice. Our study
reaction test result during hospital admissions or 21 shows how unproven drug treatments were used
days previously; therefore, patients without a coded during the first months of the pandemic, with great
diagnosis would have been excluded even if they were heterogeneity between centres, and that they were
suspected of having covid-19. The number of patients quickly replaced by proven treatments.
with covid-19 might also be underreported in clinical
settings with scarce testing resources, especially when Author affiliations
1
volumes of patients are high. In addition, medical Pharmaco- and Device Epidemiology, Centre for Statistics in
Medicine, Nuffield Department of Orthopaedics, Rheumatology, and
conditions might be underreported because the Musculoskeletal Sciences, University of Oxford, Oxford, UK
absence of a medical code for the disease is interpreted 2
Janssen Research and Development, Titusville, NJ, USA
as absence of the disease. Exposure misclassification 3
Department of Medical Informatics, Erasmus University Medical
is also possible; participating data sources varied in Center, Rotterdam, Netherlands
how drugs were captured (eg, hospital billing records, 4
Division of Cancer Sciences, School of Medical Sciences, University
prescription orders, dispensing data). of Manchester, Manchester, UK
5
Estimates for drug use on the date of hospital Nuffield Department of Orthopaedics, Rheumatology, and
Musculoskeletal Sciences, University of Oxford, Oxford, UK
admission are particularly sensitive to misclassification 6
College of Medicine and Health, University of Exeter, Exeter, UK
and could conflate baseline concomitant drug history 7
Faculty of Medicine, Islamic University of Gaza, Gaza City, Palestine
with immediate treatment on admission. We further 8
Massachusetts General Hospital, Harvard Medical School, Boston,
explored this (see supplemental figure 1) and found MA, USA
that the drugs we focused on were not typically used 9
College of Pharmacy, Riyadh Elm University, Riyadh, Saudi Arabia
before hospital admission according to the data 10
Critical Care Research Group, Nuffield Department of Clinical
sources. Neurosciences, University of Oxford, Oxford, UK
11
We did not differentiate between drugs prescribed University of Colorado Anschutz Medical Campus, Aurora, CO, USA
12
on the day of hospital admission from those in Real-World Evidence, Trial Form Support, Barcelona, Spain
13
the following days or in the context of worsening Faculty of Pharmacy, Cairo University, Cairo, Egypt
14
disease. This could also mean that some drugs used National Institute for Health and Care Excellence, London, UK
15
at discharge (or those prescribed after discharge) Regeneron Pharmaceuticals, Tarrytown, NY, US
16
could have appeared as being prescribed to patients Department of Epidemiology, Johns Hopkins Bloomberg School of
Public Health, Baltimore, MD, USA
during hospital admission. To avoid this, we censored 17
School of Population Health, UNSW Sydney, Sydney, Australia
on discharge when this information was available. 18
College of Medicine, University of Arizona, Tucson, AZ, USA
Additionally, in most of the databases where this date 19
DHC Technologies, Beijing, China
was not available, only inpatient data were provided, 20
Division of Population Health and Genomics, University of
so these drugs would not be recorded. Dundee, Dundee, UK
Another limitation of our study was the lack of 21
Department of Public Health, University of Southern Denmark,
information on dose and duration of drug treatments. Odense, Denmark
22
These are important factors that would have added School of Public Health and Community Medicine, Institute
of Medicine, Sahlgrenska Academy, University of Gothenburg,
value to our understanding of the trends in prescribing, Gothenburg, Sweden
especially among those in high risk groups or those 23
Department of Medicine, Stanford University School of Medicine,
with greater susceptibility to drug related adverse Stanford, CA, USA
events. 24
Fundació Institut Universitari per a la recerca a l’Atenció Primària
Although our study adds valuable information to de Salut Jordi Gol i Gurina (IDIAPJGol), Barcelona, Spain
25
the understanding of prescribing patterns for covid-19 Universitat Autònoma de Barcelona, Barcelona, Spain
26
in 2020, it only provides a snapshot of drug use in College of Engineering, University of Arizona Tucson, AZ, USA
27
clinical practice. As new evidence continually emerges Bayer Pharmaceuticals, Sant Joan Despí, Spain
28
over time, drug use in covid-19 is likely to evolve School of Population Medicine and Public Health, Chinese
Academy of Medical Sciences and Peking Union Medical College,
rapidly. Although possibly not representative of global Beijing, China
treatment patterns, our data provide a good oversight of 29
School of Population and Global Health, University of Melbourne,
inpatient treatment for covid-19 in real world practice Melbourne, VIC, Australia
10 doi: 10.1136/bmj.n1038 | BMJ 2021;373:n1038 | the bmjRESEARCH
30
Nanfang Hospital, Southern Medical University, Guangzhou, China Competing interests: All authors have completed the ICMJE uniform
31
Department of Biomedical Informatics, Ajou University School of disclosure form at www.icmje.org/coi_disclosure.pdf and declare:
Medicine, Suwon, South Korea support from Bill & Melinda Gates Foundation for the submitted
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
32 work; AP-U reports grants from Fundacion Alfonso Martin Escudero
VA Informatics and Computing Infrastructure, VA Salt and the Medical Research Council, outside the submitted work; AGS
Lake City Healthcare System, Salt Lake City, Utah, USA; reports personal fees from Janssen Research and Development,
Department of Internal Medicine, University of Utah School during the conduct of the study and personal fees from Janssen
of Medicine, Salt Lake City, UT, USA Research and Development, outside the submitted work; W-U-RA
33
VA Informatics and Computing Infrastructure, reports funding from the National Institute for Health Research
Tennessee Valley Healthcare System, VA Medical Center, Nashville, (NIHR) Oxford Biomedical Research Centre, Aziz Foundation,
TN, USA Wolfson Foundation, and the Royal College Surgeons of England; AG
34
Department of Biomedical Informatics, Vanderbilt University reports personal fees from Regeneron Pharmaceuticals and full time
Medical Center, Nashville, TN, USA employment at Regeneron Pharmaceuticals, outside the submitted
35 work; DRM reports funding support from the Wellcome Trust, NIHR,
Department of Preventive Medicine and Public Health, Yonsei
Scottish Chief Scientist Office, and Tenovus Scotland for research
University College of Medicine, Seoul, South Korea
36
unrelated to this work; FN was an employee of AstraZeneca until
Department of Biomedical Informatics, Columbia University, New 2019, before the conduct of this study, owns some AstraZeneca
York, NY, USA shares, and has other relationships or activities that could appear
37 to have influenced the submitted work; VS reports funding from the
IQVIA, Cambridge, MA, USA
38
Department of Biostatistics, UCLA Fielding School of Public US National Science Foundation, Agency for Healthcare Research
Health, University of California, Los Angeles, Los Angeles, CA, USA and Quality through the University of Utah, and Arizona Board of
39
Regents; DV reports personal fees from Bayer, outside the submitted
OHDSI Center at The Roux Institute, Northeastern University, work, and full time employment at Bayer; JC reports grants from the
Portland, ME, USA Korean Ministry of Health and Welfare and the Korean Ministry of
40
Columbia University Irving Medical Center, New York, NY, USA Trade, Industry, and Energy, during the conduct of the study; SCY
We thank Scott L DuVall from the Department of Veterans Affairs for reports grants from the Korean Ministry of Health and Welfare and
his contribution to this paper; the Korean Health Insurance Review the Korean Ministry of Trade, Industry, and Energy, during the conduct
and Assessment Service for providing the data; HM Hospitales for of the study; PRR reports grants from Innovative Medicines Initiative
making its data publicly available as part of the COVID Data Save Lives and Janssen Research and Development, during the conduct of the
project; and the COVID-HMAR Study Group for making Hospital del study; GH reports grants from the US National Institutes of Health
Mar data available: Juan Pablo Horcajada, Roberto Güerri, Judith Villar, (NIH) National Library of Medicine, during the conduct of the study;
Luisa Sorlí, Milagro Montero, Silvia Gómez-Zorrilla, Inmaculada López- grants from Janssen Research, outside the submitted work; CR is an
Montesinos, Mar Arenas-Miras, Joan Gómez-Junyent, Itziar Arrieta, employee of IQVIA; MAS reports grants from the US National Science
Elena Sendra, Silvia Castañeda, Emili Letang, Iván Pelegrín, Abora Rial, Foundation, US NIH, and IQVIA, personal fees from Janssen Research
Jaime Rodríguez, Carmen Gimenez, Jade Soldado, Eloi García, and and Development, during the conduct of the study; KK is an employee
Jordi Martínez Roldán. of IQVIA; PR is an employee of Janssen Research and Development
and shareholder of Johnson & Johnson; DP-A reports grants and
Contributors: All authors designed the study, interpreted the results, other from Amgen; grants, non-financial support, and other from UCB
and reviewed the manuscript. AGS designed and prepared the Biopharma; and grants from Les Laboratoires Servier, outside the
statistical package and set up the web app. AP-U, LYHL, AGS, and DP-A submitted work; and Janssen, on behalf of the Innovative Medicines
wrote the first draft of the manuscript. AP-U produced and designed Initiative (IMI) funded European Health Data and Evidence Network
the figures and tables. AP-U and AGS are joint first authors. The (EHDEN) and European Medical Information Framework (EMIF)
corresponding author attests that all listed authors meet authorship consortiums, and Synapse Management Partners have supported
criteria and that no others meeting the criteria have been omitted. AGS, training programmes organised by DP-A’s department and are open
GH, SCY, TDS, CR, KK, MG, PBR, JDP, KEL, and MM are the guarantors. for external participants.
Funding: The European Health Data and Evidence Network has Ethical approval: All the data partners received institutional review
received funding from the Innovative Medicines Initiative 2 Joint board (IRB) approval or exemption. STARR-OMOP had approval
Undertaking under grant agreement No 806968. This initiative from IRB panel #8 (RB-53248) registered to Leland Stanford Junior
receives support from the European Union’s Horizon 2020 research University under the Stanford Human Research Protection Program
and innovation programme and the European Federation of (HRPP). The use of Veterans Affairs data was reviewed by the
Pharmaceutical Industries and Associations (EFPIA). This research Department of Veterans Affairs Central IRB and was determined to
received partial support from the National Institute for Health meet the criteria for exemption under Exemption Category 4(3) and
Research (NIHR) Oxford Biomedical Research Centre, US National approved the request for waiver of the Health Insurance Portability
Institutes of Health (R01 LM00691), US Department of Veterans and Accountability Act of 1996 authorisation. The research was
Affairs, Janssen Research and Development, and IQVIA. This work approved by the Columbia University IRB as an Observational Health
was also supported by the Bio Industrial Strategic Technology Data Sciences and Informatics network study. The IRB number for use
Development programme (20001234) funded by the Ministry of of HIRA data was AJIB-MED-EXP-20-065). The use of HM Hospitales
Trade, Industry, and Energy (MOTIE, Korea) and a grant from the data as approved by the Clinical Research Ethics Committee of the
Korea Health Technology R&D Project through the Korea Health IDIAPJGol (project code: 20/070-PCV). The collection and usage of
Industry Development Institute (KHIDI), funded by the Ministry of the data for clinical research in NFHCRD was approved by the IRB of
Health and Welfare, Republic of Korea (grant No HI16C0992). This Nanfang Hospital. The use of data from the Hospital del Mar database
study was supported by the National Key Research and Development (HMAR) was approved by the Parc de Salut Mar Clinical Research
programme of China (project No 2018YFC0116901). Personal Ethics Committee.
funding included Versus Arthritis (21605), Medical Research Data sharing: Analyses were performed locally, and the patient
Council Doctoral Training Partnership (MRC-DTP) (MR/K501256/1) level data are not readily available to be shared. The analytic
(JL); MRC-DTP (MR/K501256/1, MR/N013468/1) and Fundación code is, however, available at: https://github.com/ohdsi-studies/
Alfonso Martín Escudero (FAME) (APU); Innovation Fund Denmark Covid19CharacterizationCharybdis and results are available at https://
(5153-00002B) and the Novo Nordisk Foundation (NNF14CC0001) data.ohdsi.org/Covid19CharacterizationCharybdis/
(BSKH); VINCI (VA HSR RES 13-457) (SLD, MEM, KEL); and NIHR senior
research fellowship (SRF-2018-11-ST2-004, DPA). The University of The lead authors (AP-U and AGS) affirm that the manuscript is
Oxford received funding related to this work from the Bill and Melinda an honest, accurate, and transparent account of the study being
Gates Foundation (investment ID INV-016201 and INV-019257). reported; that no important aspects of the study have been omitted;
No funders had a direct role in this study. The views and opinions and that any discrepancies from the study as planned have been
expressed are those of the authors and do not necessarily reflect explained.
those of the Clinician Scientist Award programme, NIHR, Department Dissemination to participants and related patient and public
of Veterans Affairs or the US government, the Ministry of Science communities: No public or lay dissemination activities have been
and Technology of China, and the UK National Health Service or organised to date. The Observational Health Data Sciences and
Department of Health, England. The funders had no role in considering Informatics community and contributing academic institutions, including
the study design or in the collection, analysis, interpretation of data, the University of Oxford, will issue a press release for general media and
writing of the report, or decision to submit the article for publication. the lay audience once the manuscript is available to the public.
the bmj | BMJ 2021;373:n1038 | doi: 10.1136/bmj.n1038 11RESEARCH
Provenance and peer review: Not commissioned; externally peer 23 Sanders JM, Monogue ML, Jodlowski TZ, Cutrell JB. Pharmacologic
reviewed. Treatments for Coronavirus Disease 2019 (COVID-19): A Review.
JAMA 2020;323:1824-36. doi:10.1001/jama.2020.6019
This is an Open Access article distributed in accordance with the
BMJ: first published as 10.1136/bmj.n1038 on 11 May 2021. Downloaded from http://www.bmj.com/ on 29 January 2022 by guest. Protected by copyright.
24 Mehra MR, Ruschitzka F, Patel AN. Retraction-Hydroxychloroquine
terms of the Creative Commons Attribution (CC BY 4.0) license, which or chloroquine with or without a macrolide for treatment of
permits others to distribute, remix, adapt and build upon this work, COVID-19: a multinational registry analysis. Lancet 2020;395:1820.
for commercial use, provided the original work is properly cited. See: doi:10.1016/S0140-6736(20)31324-6
http://creativecommons.org/licenses/by/4.0/. 25 Gautret P, Lagier JC, Parola P, et al. Hydroxychloroquine
and azithromycin as a treatment of COVID-19: results of an
1 Johns Hopkins University. COVID-19 Dashboard by the Center for open-label non-randomized clinical trial. Int J Antimicrob
Systems Science and Engineering at Johns Hopkins University. 2020. Agents 2020;56:105949. doi:10.1016/j.ijantimicag.2020.105949
https://coronavirus.jhu.edu/map.html. 26 Arshad S, Kilgore P, Chaudhry ZS, et al, Henry Ford COVID-19 Task
2 Wang M, Cao R, Zhang L, et al. Remdesivir and chloroquine effectively Force. Treatment with hydroxychloroquine, azithromycin, and
inhibit the recently emerged novel coronavirus (2019-nCoV) in vitro. combination in patients hospitalized with COVID-19. Int J Infect
Cell Res 2020;30:269-71. doi:10.1038/s41422-020-0282-0 Dis 2020;97:396-403. doi:10.1016/j.ijid.2020.06.099
3 US Food & Drug Administration (FDA). Coronavirus (COVID-19) 27 Horby P, Mafham M, Linsell L, et al. Effect of Hydroxychloroquine in
update: FDA evokes emergency use authorization for chloroquine Hospitalized Patients with COVID-19: Preliminary results from a multi-
and hydroxychloroquine. 2020. https://www.fda.gov/news-events/ centre, randomized, controlled trial.medRxiv [Preprint] https://www.
press-announcements/coronavirus-covid-19-update-fda-revokes- medrxiv.org/content/10.1101/2020.07.15.20151852v1.
emergency-use-authorization-chloroquine-and. 28 Github. OHDSI/CommonDataModel - Drug Era (OMOP Common
4 Horby P, Mafham M, Linsell L, et al, RECOVERY Collaborative Group. Data Model v6.0). 2018. https://www.ohdsi.org/web/wiki/doku.
Effect of Hydroxychloroquine in Hospitalized Patients with Covid-19. php?id=documentation:cdm:drug_era
N Engl J Med 2020;383:2030-40. doi:10.1056/NEJMoa2022926 29 Platt RW, Platt R, Brown JS, Henry DA, Klungel OH, Suissa S. How
5 Pan H, Peto R, Henao-Restrepo AM, et al, WHO Solidarity Trial pharmacoepidemiology networks can manage distributed analyses
Consortium. Repurposed Antiviral Drugs for Covid-19 - Interim to improve replicability and transparency and minimize bias.
WHO Solidarity Trial Results. N Engl J Med 2021;384:497-511. Pharmacoepidemiol Drug Saf 2019;29:3-7. doi:10.1002/pds.4722
doi:10.1056/NEJMoa2023184 30 World Health Organization. “Solidarity” clinical trial for COVID-19
6 Beigel JH, Tomashek KM, Dodd LE, et al, ACTT-1 Study Group treatments. 2020. www.who.int/emergencies/diseases/novel-
Members. Remdesivir for the Treatment of Covid-19 - Final Report. N coronavirus-2019/global-research-on-novel-coronavirus-2019-
Engl J Med 2020;383:1813-26. doi:10.1056/NEJMoa2007764 ncov/solidarity-clinical-trial-for-covid-19-treatments.
7 Rochwerg B, Siemieniuk RA, Agoritsas T, et al. A living WHO guideline 31 Dagens A, Sigfrid L, Cai E, et al. Scope, quality, and inclusivity of
on drugs for covid-19. BMJ 2020;370:m3379. doi:10.1136/bmj. clinical guidelines produced early in the covid-19 pandemic: rapid
m3379 review. BMJ 2020;369:m1936. doi:10.1136/bmj.m1936
8 Horby P, Lim WS, Emberson JR, et al, RECOVERY Collaborative Group. 32 RECOVERY Collaborative Group. Azithromycin in patients admitted
Dexamethasone in Hospitalized Patients with Covid-19. N Engl J to hospital with COVID-19 (RECOVERY): a randomised, controlled,
Med 2021;384:693-704. doi:10.1056/NEJMoa2021436 open-label, platform trial. Lancet 2021;397:605-12. doi:10.1016/
9 Sterne JAC, Murthy S, Diaz JV, et al, WHO Rapid Evidence Appraisal S0140-6736(21)00149-5
for COVID-19 Therapies (REACT) Working Group. Association 33 Zhang L, Lin D, Sun X, et al. Crystal structure of SARS-CoV-2 main
Between Administration of Systemic Corticosteroids and Mortality protease provides a basis for design of improved α-ketoamide
Among Critically Ill Patients With COVID-19: A Meta-analysis. inhibitors. Science 2020;368:409-12. doi:10.1126/science.
JAMA 2020;324:1330-41. doi:10.1001/jama.2020.17023 abb3405
10 Wise J. Covid-19: Arthritis drug tocilizumab reduces deaths 34 Chen FF, Zhong M, Liu Y, et al. The characteristics and outcomes of
in hospitalised patients, study shows. BMJ 2021;372:n433. 681 severe cases with COVID-19 in China. J Crit Care 2020;60:32-7.
doi:10.1136/bmj.n433 doi:10.1016/j.jcrc.2020.07.003
11 Korea covid-19 Central Clinical Task Force 2020. Antimicrobial 35 Xu P, Huang J, Fan Z, et al. Arbidol/IFN-α2b therapy for patients with
therapy recommendation. corona virus disease 2019: a retrospective multicenter cohort study.
12 Ministerio de Sanitarias (Madrid) 2020. Technical document on the Microbes Infect 2020;22:200-5. doi:10.1016/j.micinf.2020.05.012
clinical management of patients with novel coronavirus (COVID-19). 36 Shang L, Zhao J, Hu Y, Du R, Cao B. On the use of corticosteroids for
13 Centers for Disease Control and Prevention. Interim clinical guidance 2019-nCoV pneumonia. Lancet 2020;395:683-4. doi:10.1016/
for management of patients with confirmed coronavirus disease. S0140-6736(20)30361-5
2019 (COVID-19). 2020. https://www.cdc.gov/coronavirus/2019- 37 Russell CD, Millar JE, Baillie JK. Clinical evidence does not
ncov/hcp/clinical-guidance-management-patients.html. support corticosteroid treatment for 2019-nCoV lung injury.
14 World Health Organization. Clinical management of severe acute Lancet 2020;395:473-5. doi:10.1016/S0140-6736(20)30317-2
respiratory infection when novel coronavirus (nCoV) infection 38 Società Italiana di Malattie Infettive e Tropicali. 2020. [Therapeutic
is suspected: Interim guidance. 2020. https://www.who.int/ and supportive care guidelines for patients with COVID-19
publications-detail/clinical-management-of-severe-acute-respiratory- coronavirus.] http://www.fvcalabria.unicz.it/COVID-19/LINEE-GUIDA/
infection-when-novel-coronavirus-(ncov)-infection-is-suspected. linee-guida-SIMIT-marzo-2020.pdf.
15 National Health Commission of China. Diagnosis and treatment 39 Horby P, Shen Lim W, Emberson J. Effect of dexamethasone
protocols for patients with novel coronavirus pneumonia. 2020. in hospitalized patients with COVID-19: preliminary
https://www.chinadaily.com.cn/pdf/2020/1.Clinical.Protocols.for.the. report. medRxiv [preprint] 2020. www.medrxiv.org/
Diagnosis.and.Treatment.of.COVID-19.V7.pdf. content/10.1101/2020.06.22.20137273v1.
16 Dagens A, Sigfrid L, Cai E, et al. Scope, quality, and inclusivity of 40 Barnes GD, Burnett A, Allen A, et al. Thromboembolism and
clinical guidelines produced early in the covid-19 pandemic: rapid anticoagulant therapy during the COVID-19 pandemic: interim
review. BMJ 2020;369:m1936. doi:10.1136/bmj.m1936 clinical guidance from the anticoagulation forum. J Thromb
17 Voss EA, Makadia R, Matcho A, et al. Feasibility and utility Thrombolysis 2020;50:72-81. doi:10.1007/s11239-020-02138-z
of applications of the common data model to multiple, 41 Thachil J, Tang N, Gando S, et al. ISTH interim guidance on
disparate observational health databases. J Am Med Inform recognition and management of coagulopathy in COVID-19. J Thromb
Assoc 2015;22:553-64. doi:10.1093/jamia/ocu023 Haemost 2020;18:1023-6. doi:10.1111/jth.14810
18 ohdsi-studies/Covid19CharacterizationCharybdis: Charybdis v1.1.1 - 42 Bhopal R, Munro APS. Scholarly communications harmed by
Publication Package [program]. 2020. covid-19. BMJ 2021;372:n742. doi:10.1136/bmj.n742
19 Prieto-Alhambra D, Kostka K, Duarte-Salles T, et al. Unraveling 43 Subbian V, Solomonides A, Clarkson M, et al. Ethics and informatics
COVID-19: a large-scale characterization of 4.5 million COVID-19 in the age of COVID-19: challenges and recommendations for
cases using CHARYBDIS. Res Sq [preprint] 2021. doi:10.21203/ public health organization and public policy. J Am Med Inform
rs.3.rs-279400/v1 Assoc 2021;28:184-9. doi:10.1093/jamia/ocaa188
20 Datta S, Posada J, Olson G, et al. A new paradigm for accelerating
clinical data science at Stanford Medicine. 2020.
21 Rho Y, Cho DY, Son Y, et al. COVID-19 International Collaborative Web appendix: Supplementary material
Research by the Health Insurance Review and Assessment Service Using
Its Nationwide Real-world Data: Database, Outcomes, and Implications. J Web appendix: Supplementary figure 4
Prev Med Public Health 2021;54:8-16. doi:10.3961/jpmph.20.616 Web appendix: Supplementary figure 5
22 Thorlund K, Dron L, Park J, Hsu G, Forrest JI, Mills EJ. A real-
time dashboard of clinical trials for COVID-19. Lancet Digit
Web appendix: Supplementary figure 6
Health 2020;2:e286-7. doi:10.1016/S2589-7500(20)30086-8 Web appendix: Supplementary table 6
No commercial reuse: See rights and reprints http://www.bmj.com/permissions Subscribe: http://www.bmj.com/subscribeYou can also read
